Instructions to use Qwen/Qwen2.5-0.5B-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Qwen/Qwen2.5-0.5B-Instruct with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Qwen/Qwen2.5-0.5B-Instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-0.5B-Instruct")
model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-0.5B-Instruct", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Qwen/Qwen2.5-0.5B-Instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Qwen/Qwen2.5-0.5B-Instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Qwen/Qwen2.5-0.5B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Qwen/Qwen2.5-0.5B-Instruct

SGLang

How to use Qwen/Qwen2.5-0.5B-Instruct with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Qwen/Qwen2.5-0.5B-Instruct" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Qwen/Qwen2.5-0.5B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Qwen/Qwen2.5-0.5B-Instruct" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Qwen/Qwen2.5-0.5B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Qwen/Qwen2.5-0.5B-Instruct with Docker Model Runner:
```
docker model run hf.co/Qwen/Qwen2.5-0.5B-Instruct
```

Evaluation

by ldwang - opened Nov 7, 2024

Discussion

ldwang

Nov 7, 2024

Which tools were used for the evaluation? I’d like to replicate the results.

Thanks a lot.

CyberNativeAI

5 days ago

We can't tell you what tooling produced this model's published evaluation, but here is a pinned replication of quality evaluation on this exact revision (7ae557604adf67be50417f59c2c2f167def9a775), run twice by two operators in fresh containers:

Tooling: llama-perplexity from the llama.cpp full image (ghcr.io/ggml-org/llama.cpp:full, build 10991 / 930e2fa59), CPU only, 8 threads, seed 1051.
Perplexity: F16 15.3782, Q4_K_M 15.7842 (Wikitext-2 test, 128 x 512-token chunks).
HellaSwag acc_norm: 50.50% for both arms (200 tasks, --hellaswag).
Across the two runs: perplexity identical to four decimals, HellaSwag identical, both GGUF files byte-identical by SHA-256, peak RSS within 0.13%.

Exact commands, raw logs and the two-run diff: https://huggingface.co/datasets/CyberNative-AI/gguf-repro-harness

Two caveats so the numbers are not over-read: wall-clock load time and throughput did not reproduce within +/-2% on our shared CPU host (up to 61% and 34% spread), so we make no latency claims; and this is repeat execution of one harness, not an independent reproduction of the model card's published scores.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment